SOEPpaper 334: Why are East Germans not more mobile ... · Why are East Germans not more mobile?...
Transcript of SOEPpaper 334: Why are East Germans not more mobile ... · Why are East Germans not more mobile?...
Deutsches Institut für Wirtschaftsforschung
www.diw.de
Peter Boenisch • Lutz Schneider
S Why are East Germans not more mobile? Analyzing theimpact of local networks on migration intentions
334
SOEPpaperson Multidisciplinary Panel Data Research
Berlin, November 2010
SOEPpapers on Multidisciplinary Panel Data Research at DIW Berlin This series presents research findings based either directly on data from the German Socio-Economic Panel Study (SOEP) or using SOEP data as part of an internationally comparable data set (e.g. CNEF, ECHP, LIS, LWS, CHER/PACO). SOEP is a truly multidisciplinary household panel study covering a wide range of social and behavioral sciences: economics, sociology, psychology, survey methodology, econometrics and applied statistics, educational science, political science, public health, behavioral genetics, demography, geography, and sport science. The decision to publish a submission in SOEPpapers is made by a board of editors chosen by the DIW Berlin to represent the wide range of disciplines covered by SOEP. There is no external referee process and papers are either accepted or rejected without revision. Papers appear in this series as works in progress and may also appear elsewhere. They often represent preliminary studies and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be requested from the author directly. Any opinions expressed in this series are those of the author(s) and not those of DIW Berlin. Research disseminated by DIW Berlin may include views on public policy issues, but the institute itself takes no institutional policy positions. The SOEPpapers are available at http://www.diw.de/soeppapers Editors: Georg Meran (Dean DIW Graduate Center) Gert G. Wagner (Social Sciences) Joachim R. Frick (Empirical Economics) Jürgen Schupp (Sociology)
Conchita D’Ambrosio (Public Economics) Christoph Breuer (Sport Science, DIW Research Professor) Anita I. Drever (Geography) Elke Holst (Gender Studies) Martin Kroh (Political Science and Survey Methodology) Frieder R. Lang (Psychology, DIW Research Professor) Jörg-Peter Schräpler (Survey Methodology) C. Katharina Spieß (Educational Science) Martin Spieß (Survey Methodology, DIW Research Professor) ISSN: 1864-6689 (online) German Socio-Economic Panel Study (SOEP) DIW Berlin Mohrenstrasse 58 10117 Berlin, Germany Contact: Uta Rahmann | [email protected]
Why are East Germans not more mobile? Analyzing the impact of local networks on migration intentions Peter Boenischa and Lutz Schneiderb* aDepartment of Economics, Martin-Luther-University Halle-Wittenberg (MLU), Halle (Saale), Germany; bHalle Institute for Economic Research (IWH), Halle (Saale), Germany
Despite poor regional labour market conditions East Germans exhibit a rather limited willing-ness of leaving their home region. Applying an IV ordered probit approach and using the German Socio Economic Panel (SOEP), we test a local network explanation of lower spatial mobility. Firstly, we find that membership in locally bounded social networks reduces regio-nal mobility. Secondly, we show that native East Germans are more invested in this type of social networks than West Germans. Thirdly, after controlling for the social network effect the mobility gap between East and West substantially reduces. Thus, low regional labour mo-bility of East Germans is for a significant part attributable to local ties binding people to their home region. JEL classification: Z13, R23, J61. Keywords: social networks; labour mobility
* Corresponding author: Email: [email protected] Postal address: Halle Institute for
Economic Research, P.O. Box 11 03 61, D-06017 Halle (Saale), Germany
The intuition of the present analysis is that behavioural patterns of individuals in transition re-
gions are still shaped by dispositions having their seeds in the former Communist system. We
test this hypothesis by examining the impact of acculturation in a Communist regime on the
preferences for spatial mobility. We presume that people in transition regions – due to the ac-
cumulation of a system specific social capital pattern during Communism – are characterized
by strong ties to locally bounded social networks preventing them from leaving – even intend-
ing to leave – the region to an extent economic theory would expect.
In examining this presumption we combine two unconnected strands of labour market
and transition research. One the one hand a growing labour market related literature deals
with the influence of specific types of social capital on labour mobility. Following articles of
Kan (2007), Garip (2008) and David et al. (2008a) we suppose that membership in locally
bounded social networks might reduce the willingness to leave the home region. By leaving
such a bounded community – e.g. neighbourhoods or friendships – a person terminates the op-
tion of reaping returns from interactions with other members of the same network. Thus, in-
vestments in these local networks should reduce mobility. On the other hand transition litera-
ture recently verified different social capital patterns of people in Post-Communist countries
in comparison to their Western counterparts (Kaasa and Paarts 2008, Fidrmuc and Gerxani
2008, Rainer and Siedler 2009a). Whereas participation in institutionalized social capital, i.e.
membership and/or engagement in formal organisations, is underdeveloped in the East, an
abundance of informal strong tie relationships to neighbours, relatives or friends can be no-
ticed. In combining these two unrelated strands of research one is tempted to conclude that
mobility preferences of the Eastern population is weakened by its specific social capital en-
dowment.
In empirically assessing our hypothesis we focus on labour mobility of East Germans.
The reason to concentrate on East Germany is twofold. Firstly, by looking on East Germany
one is confronted with a disturbing puzzle. On the one hand, large and persistent disparities
2
between East and West German labour markets exist either in terms of unemployment or
wage rates (Aumann and Scheufele 2010). On the other hand, obstacles for migrating from
East to West are small; linguistic, institutional and spatial distance between the two parts of
Germany is negligible. Yet, regional out-migration rates in East Germany are on the same
level than in the Western part (Mai 2006). Additionally, Niebuhr et al. (2009) show that mi-
gration between Western regions is a more effective channel in equalizing regional unem-
ployment rates than in the Eastern part of Germany. With respect to labour mobility, thus,
East Germany offers a quite interesting case to test the hypothesis of Communist legacy.1
Secondly, Germany is a unique case for analysing the impact of Communism since the
differences between East and West are for the most part attributable to the recent history of
political separation after 1945 while differences between Eastern and Western Europe coun-
tries might be rooted in a large variety of historical developments in terms of culture, politics
and economics before the ‘Communist experiment’. If the Pre-Communist period affects even
the Post-Communist era (Winiecki 2004) then it is indistinct if behavioural patterns after tran-
sition are actually attributable to Communism in these countries. By contrast, focussing on
Germany offers a methodologically quite interesting option to identify the effect of Commu-
nism, as it is emphasized by the seminal work of Alesina and Fuchs-Schündeln (2007), p.
1507: “Since the political and economic system has been the same in the eastern and western
part of Germany since reunification in 1990, and was the same before 1945, West Germans
constitute a meaningful control group for East Germans.“ Therefore the German case can be
seen as natural experiment to identify “the effects of living 45 years under a Communist re-
gime on attitudes, beliefs, and political preferences”.2
Methodologically, our approach is mostly related to the labour market literature isolat-
ing the impact of local social ties on spatial mobility. The technical challenge of this kind of
analysis is the potential endogeneity of social network participation with respect to mobility.
3
The number of studies explicitly dealing with this problem is limited (Belot and Ermisch
2006, Kan 2007, David et al. 2008b). In following these studies we estimate a two equation
model of mobility preference and social network participation taking potential endogeneity of
social capital into account. Yet, in terms of content, we extend previous analyses for at least
two reasons. First, by concentrating on the case of Germany and distinguishing between ‘na-
tive’ Eastern and Western Germans we are able to identify the effect of different institutional
settings (‘Communism’ vs. ‘Liberal democracy’) on the establishment of social relationships
and, thus, on mobility preferences. Second, the German Socio-Economic Panel enables us to
measure mobility as willingness to migrate, i.e. mobility intentions. Since our hypothesis sup-
poses an enduring impact of Communism, the adequate level of analysis is the ontologically
subjective category of preferences and not actual behaviour. Because intentions are necessary
conditions for actual behaviour our study is, obviously, also relevant for explaining observ-
able mobility patterns and not solely preferences. All in all our paper, empirically, contributes
to the literature underlining the enduring impact of the Communist past on transition – an as-
pect which “in many areas of research on transition [...] tends to be underappreciated” (Win-
iecki 2008, p. 377).
Our paper opens with a theoretical chapter explicating the concept of social capital, its
relationship to spatial mobility and the impact of Communism on social capital patterns. Next,
the econometric model, the applied social capital and mobility measures as well as the identi-
fication strategy are described. Then, estimation results are displayed and discussed. A final
section draws some conclusions.
4
Social networks and spatial mobility in a Post-Communist context
The spatial dimension of social networks By analyzing the impact of social networks on economic outcomes it is usually referred to the
notion of social capital that was established in social sciences during the late 1980s and early
1990s years (Bourdieu 1986, Coleman 1988, Putnam 1993, 1995).3 Because of the young his-
tory of the concept, there is an ongoing debate on what social capital is about. Definitions
vary in being functional vs. intrinsic, normative vs. positive, individualistic vs. collectivistic.
Generally, two broad understandings of social capital can be distinguished.4 One strand – the
Bourdieu and Coleman line – refers to social capital as the investments in social networks by
individuals which provide them with resources “that they can use to achieve their interests”
(Coleman 1988, p. 101). The second strand relies on the notion of generalized trust preventing
a society from social dilemmas and promoting collective actions (Putnam 1993, Fukuyama
1995).
Our focus on the role of social networks is closely related to the first strand of the lit-
erature defining social capital as a community’s characteristic which enables its members to
reap individual returns from interactions with other members of the same community (Glaeser
et al. 2002). The distinguishing attribute of investments in social networks or social capital is
its relational structure:
“Whereas economic capital is in people’s bank accounts and human capital is inside their heads, social capital inheres in the structure of their relationships. To possess social capital, a person must be related to others, and it is these others, not himself, who are the actual source of his or her ad-vantage” (Portes 1998, p. 7).
Stressing the relational dimension of social capital is essential since it illustrates its different
nature in comparison to human capital. Leaving the network terminates the individual’s abil-
ity to gain benefits from it. Instead, the returns to human capital are less dependent on the
membership in a particular network. In our context, this aspect is crucial since it helps to ex-
plicate the spatial dimension of social capital.
5
The main characteristic of participation in social networks – investments in social cap-
ital – making it relevant for labour mobility is its dependence on a particular community.
Communities typically exhibit a geographical extension. In that sense, David et al. (2008a)
stress the localness of social networks and its implication for regional mobility. They distin-
guish two types of social networks; the first one depends on spatial proximity while the sec-
ond one is geographically unbounded. Due to this distinction the impact of participating in so-
cial networks on migration propensity is not trivial. In case of migration a membership in a
spatially bounded community runs out and the migrant’s social capital has to be depreciated.
On the other hand, spatially unbounded communities might even encourage mobility since po-
tential migrants acquire information about remote locations and easily get contact at the desti-
nation via their network connections. Thus, only a very specific type of social capital lowers
mobility.
For conceptualizing the distinction between locally bounded and unbounded networks
the theory of interpersonal ties introduced in the social network theory by Granovetter (1973)
is of great benefit. Granovetter establishes the notion of the strength of interpersonal ties:
“Most intuitive notions of the ‘strength’ of an interpersonal tie should be satisfied by the following definition: The strength of a tie is a (probably linear) combination of the amount of time, the emo-tional intensity, the intimacy (mutual confiding), and the reciprocal services which characterize the tie” (Granovetter 1973, p. 1361).
Our hypothesis relating the strength of ties to spatial mobility states that locally bounded net-
works operate on the basis of strong ties, i.e. regularly and intense personal contacts between
specific individuals. At least to some extent, these contacts require spatial proximity between
particular individuals. One might consider a few special cases where regularly contact via
spatial proximity can be partly substituted by media. Nevertheless, the basic kind of establish-
ing a strong tie network and building up reputation between participants is due to face-to-face
interactions. Therefore, the geographic extension of such networks is limited. Furthermore,
strong tie networks can be characterized as closed communities. Information generated within
6
the network circulate very fast but the capability to acquire credible information from outside
the network is very limited. Therefore, the recognition of outside opportunities encouraging
mobility is reduced.
On contrary, networks operating on the basis of weak ties, i.e. less frequent and in-
tense personal contacts, are able to transcend spatial boundaries. Networks of this type exhibit
a rather open character, hence, information on opportunities in distant regions can be acquired
via weak ties. Granovetter states, that “whatever is to be diffused can reach a larger number of
people, and traverse greater social distance [...], when passed through weak ties rather than
strong” (Granovetter 1973, p. 1366) Furthermore, in case of migration, the accumulated social
capital keeps its economic value since weak ties to members in the host region can be turned
into strong ties after migration. Therefore, participation in such networks is less tied to a cer-
tain location and, as a consequence, does not reduce mobility or actually foster it. Belonging
to a close knit exclusionary network of strong ties, on the other hand, should prevent partici-
pant from moving to other regions. Otherwise their accumulated network capital, for the most
part, would be useless and has to be depreciated.
Another aspect has to be taken into consideration. If participation in social networks
characterized by strong ties affects mobility, then, individuals seriously considering to move
away should adjust their investment behaviour. Individuals with strong mobility preferences
should invest less in locally bounded network activities while immobile people might prefer
these strong relationships to a locally concentrated community. In other words, membership
in social networks exerts influence on mobility but vice versa mobility intentions should in-
fluence the network activities. In the empirical analysis, this interdependent relationship be-
tween social interaction and mobility has to be taken into account otherwise a simultaneity bi-
as arises.
7
Social networks under the totalitarian rule Our analysis crucially rests upon the hypothesis that East and West Germans differ in their
social network patterns. East Germans, we suppose, are more connected to locally bounded
networks and, thus, show a rather limited spatial mobility. The reason behind the hypothesis
of the East Germans’ localness is the acculturation in a totalitarian political system and the
following abrupt institutional transformation. A multifaceted literature deals with the impact
of totalitarian – particularly the Communist – systems and the following transformation pe-
riod on social capital investments (Mihailova 2005). One motive for the extensive debate on
this issue is, according to Paldam and Svendsen (2002), the conjecture that social capital acts
as missing link in explaining the slow adjustment of economic and social domains in Eastern
countries to standards of developed western countries.5
The underlying argument is based on the recognition of a social capital gap between
eastern and western economies recently confirmed by the analysis of Fidrmuc and Gërxani
(2008). However, most authors agree that the lack of social capital considers only the institu-
tionalized type of social capital which is built up within the legal framework (Mihailova
2005) as well as the dimension of generalized social trust measuring the trust to people not
belonging to the own close-knit community (Rainer and Siedler 2009b). On the contrary, net-
works of families, friends or kinship based on strong ties seem to play a more important role
than in the western countries.6 Rose (1999) classifies this type of social capital as ‘negative’
or ‘anti-modern’ since it acts as obstacle to institutional transformation, i.e. the actual en-
forcement of the rule of law. The reason for a different social capital pattern in transition
countries is twofold. Firstly, according to the so called dictatorship theory the former totalitar-
ian system destroyed civic participation and trust in formal institutions and caused a retreat in-
to closed informal networks:
“All Communist countries had experienced a phase of stark, totalitarian rule; and even after severe repression ended with the Stalinist era, participation in public affairs remained forced and ritualis-
8
tic. People therefore tended to retreat from the public sphere into privacy; into the realm of rela-tives and immediate friends; or into innocuous groups promoting non-controversial cultural and leisure activities. Public institutions were perceived as [...] imposed by a foreign power” (Raiser 2001, p. 4).
Hence, “under the communist system an autonomous ‘social tissue’ was destroyed” (Mickie-
wicz 2009, p. 404). Secondly, after the breakdown of the totalitarian rule an institutional vac-
uum occurred and the informal networks became even more necessary to cope with the risks
of the transition period. Keeping this line of reasoning in mind one would suggest that spatial
mobility is rather limited in transition economies. If individuals believe that they only can
“profit from informal social capital returns” (Raiser 2001, p. 4) they will not jeopardize these
relationships by leaving the community. However, regarding the special case of East Ger-
many a somewhat different development was observed. According to Rainer and Siedler
(2009b) trust in institutions rapidly regenerated during the transition period. Due to the imme-
diate takeover of political and legal institutions from West Germany an institutional vacuum
was prevented and, as a consequence of the performance of the imported system, institutional
trust was renewed.
All in all, the social capital pattern of East Germans is supposed to be dominated by
strong tie relationships of informal networks in contradiction to weak tie oriented open net-
works; yet, trust in impersonal institutions seems to be established quite well. Characterized
by this social capital pattern East Germans are more locally bounded and show rather limited
willingness of leaving their home region. Turning to the empirical part of the analysis we de-
rive the following three hypotheses:
1) Participation in social networks characterized by strong ties discourages mobility (Mobility hypothesis).
2) Native East Germans are more related to strong tie social networks than West Ger-mans (East hypothesis).
3) Controlling the social network effect substantially reduces the mobility gap between East and West Germans (Gap hypothesis).
9
Empirical analysis
Econometric model As outlined in the previous section there might be a simultaneous dependency between the in-
dividual social network pattern *iC and the individual mobility proneness *
iM . We estimate
this interdependent relationship using the Amemiya Generalized Least Square Estimator
(Amemiya 1979) and find no evidence for the impact of mobility propensity on the social
network structure.7 Hence, in line with the work of Kan (2007) and Belot and Ermisch (2006)
the following recursive two equation model turns out to be appropriate:
(1) iiii xCM 11** ' εβγ ++=
(2) iiii zxC 22* '' επβ ++=
where 21,ββ and π are vectors of parameters, ix is a vector of socioeconomic variables and
ii 21 ,εε are normally distributed errors (allowed to be correlated) with a zero mean ( )Σ,0N . In
order to ensure identification the second equation of the model contains a set of instrumental
variables iz discussed below. Furthermore we perform a likelihood ratio test to avoid biased
estimation results due to weak instruments.
Instead of the latent tendencies *iM and *
iC we observe the ordinal variables iM and iC such that:
(3)
⎪⎪⎩
⎪⎪⎨
⎧
<≤<≤<
≤
=⎪⎩
⎪⎨
⎧
<≤<
≤=
*23
23*
22
22*
21
21*
*12
12*
11
11*
if3 if2 if1 if0
if2 if1 if0
i
i
i
i
i
i
i
i
i
CCC
C
CMM
MM
μμμμμ
μ
μμμ
μ
The unknown cutoffs and parameters are efficiently estimates by Full Information Maximum
Likelihood (FIML).
10
Data and measurement In our inquiry, we use the German Socio-Economic Panel (SOEP), a representative survey of
German households (Wagner, Frick and Schupp 2007). The data set enables us to measure the
social relationships and provides information on the mobility intentions. Since only some
waves contain the relevant records on social networks activities and mobility intentions we
have to focus on the wave of 1999. Due to the dependency of mobility decisions within
households we only use information on individuals that are classified as household heads.
Furthermore we restrict our analysis to people relevant for the labour market aged between 16
and 65 years. Because of these restrictions and some missing value problems the sample size
of our analysis reduces to almost 3,600 individuals.
The crucial question for the analysis considers the measurement of the endogenous
variables of an individual at a certain point in time. We quantify the individual social net-
work activities via the frequency of helping friends, relatives or neighbours
(HELPFRIENDS). Given the reciprocity of strong ties mentioned by Granovetter (1973) this
variable approximates the amount of locally bounded social networks a person has access to.
The variable stems from the following question:
HELPFRIENDS: “Please indicate how often you take part in each activity: every week, every month, rarely or nev-er?” “Lend help to friends, relatives or neighbours when something has to be done” (SOEP variable code PP0305)
The variable is ordinarily coded; the higher the value the more social capital in terms of po-
tential assistance in the future from friends, relatives or neighbours is acquired.
To assess an individual’s mobility preference at the date when social network activi-
ties are measured we rely on the SOEP question whether the person considers moving away.
The question is expressed in a way that should exclude nearby moves within a region. Thus,
the question precisely measures that type of mobility relevant for locally bounded social net-
work activities.
11
MOVE_INTENT: “Would you consider moving away, e.g. because of family or job?” “Yes; possibly, can't exclude the possibility; no” (SOEP variable code PP114)
There are several aspects to note about the mobility measure applied in this inquiry. In con-
trast to the usually applied proxies based on actual moves, stated mobility preferences or in-
tentions have three advantages. Firstly, individual behaviour is guided by this ontologically
subjective category. People should reduce their local ties if they intend to move away even if
they actually do not move. Secondly, in the German formulation the question is expressed in a
way which is directly related to social networks. Moving away exactly means cutting local
ties. If, conversely, actual moves are used one has to determine the spatial dimension of social
networks by herself. By applying the stated preference variable this decision is left to the in-
terviewee. Finally, because of panel attrition employing actual moves is likely to induce a se-
lectivity bias which is difficult to account for in the context of endogenous regressors. All
previous studies ignore this problem. Table 1 describes the endogenous variables and the way
they are operationalized.
Table 1. Description of endogenous variables Variable Description Measurement
MOVE_INTENT Mobility intention Expressed mobility intention within next two years (0 = no intention; 1 = maybe; 2 = yes)
HELPFRIENDS Social Network Frequency of helping friends (0 = never ... 3 = weekly)
For testing the east hypothesis (H2) a variable is required representing the institutional regime
in which a person is grown up. We include a variable called EASTORIGIN that is based on the
following question:
EASTORIGIN: “Where did you live before German reunification, i.e. before 1989?” “GDR (including East Berlin), Federal Republic (including West Berlin), Abroad” (SOEP variable code TP121)
12
So, persons that moved from East Germany to the Western part after the Fall of the
Wall are identified as acculturated in a Communist institutional setting. Hence, the potential
selectivity bias due to east-west migration is reduced to the negligible amount of moves be-
fore 1989. Moreover, to assure the validity of the natural experiment of the German separa-
tion we exclude all foreigners and Germans who lived abroad before 1989. Because of the
almost exogenous character of the EASTORIGIN variable determined in this way we imple-
ment it as independent regressor in both equations of the model.
The exogenous variable used in the estimation are depicted in table 2. They include
the main personal characteristics usually applied in the analysis of determinants of mobility
and social capital, i.e. age, sex, family status, education, employment status, household in-
come (Kan 2007, Belot and Ermisch 2006).
13
Table 2: Description of exogenous variables
Variable Description Measurement
EASTORIGIN Lived 1989 in East Germany 1=yes, 0=no
AGE Age Age in years
FEMALE Female 1=yes, 0=no
MARRIED Marital status: Married, living together 1=yes, 0=no
SEPARATED Marital status: Married, living separated 1=yes, 0=no
SINGLE Marital status: Single 1=yes, 0=no
DIVORCED Marital status: Divorced 1=yes, 0=no
WIDOWED Marital status: Widowed 1=yes, 0=no
CHILD Number of children Number of children under 17 years living in Household
FAMILY_ CHANGE
Household composition change last year 1=yes, 0=no
EDUCATION Education Duration of education in years
EMPLOYED Employment status 1=full-time employed, 0=other
INCOME Household income Monthly net household income in Euro (after taxes and transfers)
PROB_ UNEMPLOY
Subjective unemployment risk
0% to 100 % risk estimation of losing current job within two years
FLAT_OWNER Flat owner Owner of the flat where household lives (1=yes, 0=no)
CARE Person in household needing constant care 1=yes, 0=no
FINISH_EDUC Completion of education last year 1=yes, 0=no
14
Identification As discussed above the IV approach conducted in this paper requires a set of certain instru-
mental variables to ensure proper identification. Valid Instrumentation rests upon two major
presuppositions. Firstly, instruments should be (highly) correlated with the endogenous, i.e.
the HELPFRIENDS variable. Secondly, instruments have to be uncorrelated with – orthogo-
nal to – the error term in the structural, i.e. the MOVE_INTENT equation. Whereas the first
condition can be assessed via a weak instruments test the second criteria must be verified by
theoretical considerations since the usual tests of overidentification restrictions are not feasi-
ble in case of ordered probit models. In our analysis we use four instruments displayed in Ta-
ble 3.
Table 3: Description of Instruments
Regarding the first premise we have to verify the relevance of the four variables for our social
capital measure. From the social network analysis point of view (Granovetter 1973) the num-
ber of siblings (SIBLINGS) should increase contact potential since the social ties of siblings
can be used to form own ties. Thus, the number of siblings increases an individual’s contact
pool. Additionally, we implement a variable reflecting the educational background of the fa-
ther (FATHER_TRAIN) since this kind of variable is commonly used as indicator for the so-
ciability of persons. Thirdly, we include the duration the person is living in her flat
(FLAT_DURATION) since a high duration should be highly correlated to neighbourhood con-
tacts and friendship help. Finally, we implement the FIRM_TENURE variable because a high
Variable Description Measurement
SIBLINGS Number of siblings
FATHER_TRAINING Father with high level of vocational training 1=yes, 0=no
FLAT_DURATION Duration of living in flat Duration in years
FIRM TENURE Firm Tenure Tenure in years
15
tenure should increase the opportunity of building up strong informal ties (e.g. to colleagues).
As one can see in the next section the performed weak instruments test rejects the null hy-
pothesis of no relevance of instruments for the HELPFRIENDS variable. Thus, it is justified
to conclude that the first requirement for correct identification is met.
A more critical aspect concerns the second condition for valid instrumentation, i.e. un-
correlatedness or orthogonality of instruments with the error term in the MOVE_INTENT
equation. Before discussing the itemized variables one should conceive that valid identifica-
tion of course allows an instrument’s influence on intentions to move via the endogenous var-
iable HELPFRIENDS or due to its correlation to other included exogenous variables. If, for
example, an individual’s firm tenure increases its income and reduces, therefore, mobility no
correlation between firm tenure and the error term of the mobility equation occurs and the or-
thogonality condition is satisfied.
Regarding SIBLINGS we do not find any provable impact on the intention to move
away. One might argue that having siblings exerts influence on mobility preferences due to
care considerations with respect to parents (Rainer and Siedler 2009c). Yet, we control for
this aspect by implementing the CARE variable. With respect to FATHER_TRAINING a heri-
tage effect seems to be likely, i.e. children of trained fathers are more mobile since they are
also more trained. However, after taking the own educational level into account this effect
does not lead to a correlation between the instrument and the error term. Likewise, we sup-
pose that when FLAT_DURATION increases the intention to move clearly reduces. Neverthe-
less, the orthogonality condition seems to be met since it is the growing social network –
measured by our endogenous variable – that weakens mobility and not duration of living in a
flat as such. Considering FIRM_TENURE we would like to claim that the impact on mobility
intentions works through i) social capital and ii) wages. Obviously, firm tenure reduces job
and spatial mobility due to the beneficial embeddedness of a worker in the firm’s social net-
16
work. Furthermore, tenure and wages are positively correlated. After moving away wage
components solely based on seniority – i.e. Lazear’s (1981) deferred payment wage scheme –
are lost. Therefore higher FIRM_TENURE should be associated with lower mobility prefer-
ence. Yet, we control both effects by the HELPFRIENDS and INCOME variable. Hence we
conclude, from a theoretical point of view, the orthogonality condition of appropriate instru-
mentation seems to be satisfied by our identification strategy.
Results For testing our hypotheses (1)-(3) we estimate the bivariate ordered probit mobility model
with an endogenous regressor for social networks as explicated in chapter three. However, to
test hypothesis (3) we also run a ‘naive’ ordered probit regression of the mobility model ne-
glecting the effect of social networks. By comparing the adequate and the ‘naive’ model we
are able to evaluate the social network effect of East Germans on their mobility intentions.
Table 4 displays the results for the adequate mobility model taken the endogenous na-
ture of the social network variable into account. The left column contains the results for the
mobility equation (1). The right column displays the instrumented estimation for the social
network equation (2). The model seems to be well specified as, firstly, can be seen from the
Wald-Test of overall significance and, secondly, from the joint significance of instruments
used to identify the social network effect.8
17
Table 4: Estimation results I (Full Information Maximum Likelihood Estimation)
Mobility
[MOVE_INTENT] (Equation 1)
Social Capital
[HELPFRIENDS] (Equation 2)
coefficient sd. coefficient sd.
Endogenous Variables HELPFRIENDS -0.9373*** 0.0500 - -Exogenous Variables EASTORIGIN -0.0329 0.0587 0.0869 ** 0.0410FEMALE -0.1687*** 0.0418 -0.1452 *** 0.0412AGE -0.0385*** 0.0147 -0.0470 *** 0.0140AGE2 0.0002 0.0002 0.0003 ** 0.0002MARRIED 0.0986 0.0620 0.1222 ** 0.0601SEPARATED 0.0603 0.1214 -0.0048 0.1216DIVORCED -0.0044 0.0751 -0.0083 0.0737WIDOWED 0.2397** 0.1022 0.2651 *** 0.0982CHILD -0.0858*** 0.0268 -0.0634 ** 0.0261FAMILY_CHANGE 0.0163 0.0531 -0.0012 0.0509EDUCATION 0.0219 0.0136 -0.0088 0.0080INCOME -0.0000* 0.0000 -0.0000 *** 0.0000UNEMPLOYED 0.0140 0.0560 -0.0361 0.0538PROB_UNEMPLOY 0.0012 0.0009 0.0002 0.0008FLAT_OWNER -0.1036 0.0771 0.0631 0.0436CARE -0.2172** 0.1080 -0.2008 * 0.1035FINISH_EDUC 0.0311 0.1315 -0.0668 0.1229FLAT_DURATION - - 0.0056 *** 0.0019FIRM TENURE - - -0.0007 0.0010SIBLINGS - - 0.0045 0.0049FATHER_TRAINING - - -0.0332 0.0215
Wald statistic p-value
OVERALL SIGNIFICANCE 141.18 (21) *** 0.0000
WEAK INSTRUMENTS TEST 8.88 (4) * 0.0643
NO. OBSERVATIONS 3627
Notes: Number of hypotheses tested is given in parentheses. Significance levels are 1% (***), 5% (**) and 10% (*). Cutoffs
are not displayed. The constant is restricted to zero.
Source: SOEP 1999.
18
Before discussing the variables of primary interest we briefly inspect the controls. In
general, results are in line with previous empirical research. We find significant effects of
gender, age, marital status, children, income, and home care obligations on mobility (table 4,
left column). Surprisingly, educational level, (un-)employment aspects, and flat ownership do
not seem to play a major role for mobility preference. With respect to the instrumental equa-
tion of participating in strong tie networks we confirm a significant impact of gender, age,
marital status, children, income, and home care on the intensity of friendship relations (table
4, right column). In addition, flat duration has a verifiable effect on the social relationship var-
iable and, via this channel, on mobility preference as well. Somewhat unexpected, the educa-
tion and employment variables still seem to be of minor importance for explaining the
strength of social relationships and/or mobility considerations. However, one should be cau-
tious in drawing heavy conclusions from these results since multicollinearity problems regard-
ing these variables could lead to low statistical inference.
Turning to our hypotheses, in the left column of table four we find clear evidence for
the mobility hypothesis (H1). Our measure of joining a social network characterized by strong
ties – i.e. the variable representing the frequency of helping friends – shows a significant neg-
ative parameter estimate. Thus, being member in a strong tie social network significantly re-
duces the willingness to move away. With respect to the hypothesis (H2) – the impact of ac-
culturation in the Eastern part of Germany on the individual social network pattern – we find
clear support for our conjecture. After controlling for individual characteristics native East
Germans are more invested in strong tie relationships than West Germans. Taking together
hypotheses (1) and (2) we are justified to conclude that the specific pattern of East Germans’
relationships weakens the willingness of leaving home. In line with this implication the dum-
my variable representing the effect of being a native East German (Eastorigin) does not seem
to have a verifiable impact on mobility preferences.
19
Table 5: Estimation results II (Ordered Probit Estimation)
Mobility
[MOVE_INTENT] (Naive model structure)
coefficient sd.
Exogenous Variables EASTORIGIN -0.2830*** 0.0423 FEMALE -0.0783* 0.0425 AGE 0.0118 0.0148 AGE2 -0.0004** 0.0002 MARRIED -0.0504 0.0624 SEPARATED 0.1897 0.1235 DIVORCED 0.0401 0.0795 WIDOWED -0.0254 0.1236 CHILD -0.0616** 0.0255 FAMILY_CHANGE 0.0501 0.0519 EDUCATION 0.0805*** 0.0082 INCOME 0.0000*** 0.0000 UNEMPLOYED 0.1060* 0.0573 PROB_UNEMPLOY 0.0025*** 0.0009 FLAT_OWNER -0.4467*** 0.0439 CARE -0.0887 0.1403 FINISH_EDUC 0.2317* 0.1209
Wald statistic p-value OVERALL SIGNIFICANCE 597.774(17) *** 0.0000
NO. OBSERVATIONS 3627
Notes: Number of hypotheses tested is given in parentheses. Significance levels are 1% (***), 5% (**) and 10% (*). Cutoffs
are not displayed. The constant is restricted to zero.
Source: SOEP 1999.
To confirm our explanation of the potential mobility gap of East Germans we estimate a ‘na-
ive’ model ignoring the impact of social networks. By estimating the 'naive' model we, fur-
thermore, are able to the test the sensitivity of the preferred structural mobility model. In table
five the ordered probit regression of the mobility model neglecting the endogenous network
regressor is shown. We estimate the same model as in table four but omit the variable measur-
20
ing the frequency of helping friends. Results are in accordance with our gap hypothesis (H3).
If the effect of the special pattern of East Germans’ social relationships is ignored in the mo-
bility equation the East dummy becomes significant negative. So we find strong evidence for
(H3) meaning that the mobility gap disappears after controlling the social network effect. To a
substantial part, East Germans are less mobile than West Germans due to the effect of their
specific ties to local networks. To put it more generally, our analysis confirms a significant
negative effect of being acculturated in the East on mobility preference.
Conclusion Despite considerable and persistent labour market differences between East and West spatial
mobility (preference) of East Germans is rather limited. Our analysis focuses on a ‘Commu-
nist legacy’ explanation of moderate labour mobility. We hypothesize that acculturation in a
totalitarian system led to a social capital pattern characterized by strong ties to locally
bounded networks causing a lower willingness to leave the home region.
Our results are in favour of this conjecture. By using des German Socio Economic
Panel and estimating a bivariate ordered probit model with an endogenous regressor we firstly
find significant differences between networks East and West Germans are joining. East Ger-
mans are more invested in informal social networks characterized by local ties (East hypothe-
sis). Second, we show that such informal and strong relationships significantly reduce spatial
mobility (Mobility hypothesis). Furthermore, a comparison of our mobility model controlling
for social network participation in an appropriate way and a ‘naive’ model neglecting the in-
fluence of networks reveals that the mobility gap between East and West Germans disappears
if the social network effect is taken into account (Gap hypothesis).
Altogether, we conclude that acculturation in the Communist system contributes to
explain different social network structures and, as a consequence, different behavioural pat-
terns and mental dispositions in terms of mobility. During Communism East Germans built up
21
strong ties to locations where labour market opportunities radically altered and often de-
creased after the political and economic breakdown. The price a lot of them had to pay was
low labour market performance during transition – unless the unpleasant option of moving
away and terminating strong social relationships was chosen.
Notes
1. Even if we focus on Germany the mobility topic has a similar relevance for a number of Post-Communist
countries. Fidrmuc (2004), Gács and Huber (2005) as well as Bornhorst and Commander (2006) point to
the low level of regional adjustments to labour market disparities and shocks via the channel of worker re-
location in these countries. Paci et al. (2007) confirm the very strong attachment of individuals in the new
EU member states to their local community.
2. See also Rainer and Siedler who reflect the literature on the natural experiment character of the recent
German history and use „the German separation and reunification as an exogenous event to estimate the
causal effect of communism“ (Rainer and Siedler 2009b, p. 255).
3. For the economic approaches to social capital see the review articles of Paldam (2000), Sobel (2002), Dur-
lauf and Fafchamps (2005) and Dasgupta (2005).
4. Paldam (2000) distinguishes three conceptual families: trust, cooperation and networks. Nevertheless, as
Paldam himself admits, the cooperation and the trust concept are very similar; thus, they might be unified
to one category.
5. A related discussion deals with the effects of social capital in developing countries (Dasgupta 2005). How-
ever, as Winiecki (2004) noted, the observed lack of ‘civilisational fundamentals’ of liberty, law and order,
and trust might reach back to the Pre-Communist era.
6. Ledeneva (1998) shows the abundant social capital in terms of informal networks and so called blat rela-
tionships for Russia.
7. In order to test the simultaneous structure we extend equation 2 given above with the endogenous regressor
*iM and the corresponding parameter φ : iiiii zxMC 11
** '' επβφ +++= Afterwards, on the first stage, we
simultaneously estimate the reduced form of this modified two equation ordered probit model via Maxi-
mum Likelihood. On the second stage, the structural parameters are estimated by Generalized Least Square
method on the basis of the coefficients of the first stage. The corrected covariance matrix is calculated ac-
22
cording to Amemiya (1979). This two stage procedure gives consistent, but still inefficient parameter esti-
mates. Since we cannot reject the hypothesis 0=φ we conduct the efficient Full Information Maximum
Likelihood procedure described below.
8. However, beside the performed weak instruments test it would be useful to have a test for the orthogonality
condition i.e. the uncorrelatedness of the instruments and the error term of the structural equation. In the
case of an ordered probit model the common tests of overidentifying restrictions are not feasible.
References Alesina, A., and Fuchs-Schündeln, N., 2007. ‘Good-Bye Lenin (or Not?): The Effect of
Communism on People's Preferences’. The American Economic Review, 97 (4),
1507–28.
Amemiya, T., 1979. The Estimation of a Simultaneous-Equation Tobit Model. International
Economic Review, 20 (1), 169–181.
Aumann, B., and Scheufele, R., 2010. Is East Germany catching up? A time series perspec-
tive. Post-Communist Economies, 22 (2), 177–192 (forthcoming).
Belot, M., and Ermisch, J., 2006. Friendship Ties and Geographical Mobility: Evidence from
the BHPS. IZA Discussion Paper 2209. Institute for the Study of Labor (IZA).
Bornhorst, F., and Commander, S., 2006. Regional unemployment and its persistence in tran-
sition countries. Economics of Transition, 14 (2), 269–288.
Bourdieu, P., 1986. The Forms of capital. In: J.G. Richardson, ed. Handbook of theory and
research for the sociology of education. New York: Greenwood, 241–258.
Coleman, J.S., 1988. Social capital in creation of human capital. American Journal of Sociol-
ogy, 94 (Supplement), 95–120.
Dasgupta, P., 2005. Economics of Social Capital. The Economic Record, 81 (s1), S2–S21.
23
David, Q., Janiak, A., and Wasmer, E., 2008a. Local Social Capital and Geographical Mobil-
ity: A Theory. IZA Discussion Paper 3668. Institute for the Study of Labor (IZA).
David, Q., Janiak, A., and Wasmer, E., 2008b. Local Social Capital and Geographical Mobil-
ity: Some Empirics and a Conjecture on the Nature of European Unemployment.
IZA Discussion Paper 3669. Institute for the Study of Labor (IZA).
Durlauf, S.N., and Fafchamps, M., 2005. Social Capital. In: P. Aghion and S. Durlauf, eds.
Handbook of Economic Growth. 1st ed. Volume 1. Amsterdam: Elsevier, 1639–
1699
Fidrmuc, J., 2004. Migration and regional adjustment to asymmetric shocks in transition
economies. Journal of Comparative Economics, 32 (2), 230–247.
Fidrmuc, J., and Gërxhani, K., 2008. Mind the gap! Social capital, East and West. Journal of
Comparative Economics, 36 (2), 264–286.
Fukuyama, F., 1995. Trust: the social virtues and the creation of prosperity. New York: The
Free Press.
Gács, V., and Huber, P., 2005. Quantity adjustments in the regional labour markets of EU
candidate countries. Papers in Regional Science, 84 (4), 553–574.
Garip, F., 2008. Social capital and migration: how do similar resources lead to divergent out-
comes? Demography, 45 (3), 591–617.
Glaeser, E.L., Laibson, D., and Sacerdote, B., 2002. An Economic Approach to Social Capi-
tal. Economics Journal, 112, F437–F458.
Granovetter, M., 1973. The strength of weak ties. American Journal of Sociology, 78,1360–
80.
Kan, K., 2007. Residential Mobility and Social Capital. Journal of Urban Economics, 61 (3),
436–457.
24
Kaasa, A. and Parts, E., 2008. Individual-level determinants of social capital in Europe: dif-
ferences between country groups. Acta sociologica, 51 (2), 145–168.
Lazear, E. P., 1981. Agency, earnings profiles, productivity, and hours restrictions. The Amer-
ican Economic Review, 71 (4), 606–620.
Ledeneva, A.V., 1998. Russia’s economy of favours. Blat, Networking and Informal ex-
change. Cambridge: Cambridge University Press.
Mai, R., 2006. Die altersselektive Abwanderung aus Ostdeutschland. Raumforschung und
Raumordnung, 5, 355–69.
Mickiewicz, T.M., 2009. Hierarchy of governance institutions and the pecking order of priva-
tisation: Central-Eastern Europe and Central Asia reconsidered. Post-Communist
Economies, 21 (4), 399–423.
Mihaylova, D., 2005. Social Capital in Central and Eastern Europe: A Critical Assessment
and Literature Review. EconWPA Papers, Development and Comp Systems
0511001.
Niebuhr, A., Granato, N., Haas, A., and Hamann, S., 2009. Does labour mobility reduce dis-
parities between regional labour markets in Germany? IAB Discussion Paper
200915, Nuremberg, IAB.
Paci, P., Tiongson, E., Walewski, M., Liwinski, J., and Stoikova, M.M., 2007. Internal Labor
Mobility in Central European and the Baltic Region. World Bank Working Paper
105.
Paldam, M., 2000. Social Capital: One or Many? Definition and Measurement. Journal of
Economic Surveys, 14 (5), 629–53.
25
Paldam, M., and Svendsen, G.T., 2002. Missing Social Capital and the Transition from So-
cialism. Journal for Institutional Innovation, Development and Transition, 5, 21–
34.
Portes, A., 1998. Social Capital: Its Origins and Applications in Modern Sociology. Annual
Review of Sociology, 24, 1–24.
Putnam, R.D., 1993. Making Democracy Work: Civic Traditions in Modern Italy. Princeton:
Princeton University Press.
Putnam, R.D., 1995. Bowling alone: America's declining social capital. Journal of Democ-
racy, 6 (1), 65–78.
Rainer, H., and Siedler, T., 2009a. The role of social networks in determining migration and
labour market outcomes. Economics of Transition, 17 (4), 739–67.
Rainer, H., and Siedler, T., 2009b. Does democracy foster trust? Journal of Comparative
Economics, 37 (2), 251–69.
Rainer, H., and Siedler, T., 2009c. O Brother, Where Art Thou? The Effects of Having a Sib-
ling on Geographic Mobility and Labour Market Outcomes. Economica, 76 (303),
528–556.
Raiser, M., Haerpfer, C., Noworthy, T., and Wallace, C., 2001. Social Capital in transition: a
first look at the evidence. European Bank for Reconstruction and Development
Working Paper 61.
Rose, R., 1999. Modern, pre-modern and anti-modern social capital in Russia. Studies in Pub-
lic Policy Working Paper 324, Centre for the Study of Public Policy, University of
Strathclyde, Glasgow.
Sobel, J., 2002. Can We Trust Social Capital? Journal of Economic Literature, 40 (1), 139–
154.
26
Wagner, G. G., Frick, J. R., and Schupp, J., 2007. The German Socio-Economic Panel Study
(SOEP) - Scope, Evolution and Enhancements. Schmollers Jahrbuch 127 ( 1),
139-169.
Winiecki, J., 2004. Determinants of catching up or falling behind: interaction of formal and
informal institutions. Post-Communist Economies 16 (2), 137–152.
Winiecki, J., 2008. Employment and unemployment in transition: the legacy of the commu-
nist past. Post-Communist Economies 20 (3), 377–390.